The Profits of Power: Land Rights and Agricultural ...

The Profits of Power: Land Rights and Agricultural Investment in GhanaAuthor(s): Markus Goldstein and Christopher UdrySource: Journal of Political Economy, Vol. 116, No. 6 (December 2008), pp. 981-1022Published by: The University of Chicago PressStable URL: http://www.jstor.org/stable/10.1086/595561 .Accessed: 01/12/2015 09:07

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Journalof Political Economy.

http://www.jstor.org

This content downloaded from 23.235.32.0 on Tue, 1 Dec 2015 09:07:14 AMAll use subject to JSTOR Terms and Conditions

981

[ Journal of Political Economy, 2008, vol. 116, no. 6]! 2008 by The University of Chicago. All rights reserved. 0022-3808/2008/11606-0001$10.00

The Profits of Power: Land Rights andAgricultural Investment in Ghana

Markus GoldsteinWorld Bank

Christopher UdryYale University

We examine the impact of ambiguous and contested land rights oninvestment and productivity in agriculture in Akwapim, Ghana. Weshow that individuals who hold powerful positions in a local politicalhierarchy have more secure tenure rights and that as a consequencethey invest more in land fertility and have substantially higher output.The intensity of investments on different plots cultivated by a givenindividual corresponds to that individual’s security of tenure overthose specific plots and, in turn, to the individual’s position in thepolitical hierarchy relevant to those specific plots.

I. Introduction

Institutions matter for growth and development. In particular, it is ap-parent that investment incentives depend on expectations of rights over

This research has received financial support from the National Science Foundation(grants SBR-9617694 and SES-9905720), International Food Policy Research Institute,Institute for the Study of World Politics, World Bank Research Committee, FulbrightProgram, Economic and Social Research Council, and the Pew Charitable Trust. We thankDaron Acemoglu, Ernest Appiah, Oriana Bandiera, Abhijit Banerjee, Xu Cheng, EstherDuflo, Michael Kremer, Eliana La Ferrara, W. Bentley MacLeod, Roger Myerson, HowardPack, Rohini Pande, Ashok Rai, Tony Smith, Duncan Thomas, and Justin Wolfers for veryhelpful comments. Research for this paper was conducted while both authors were visitingscholars at the Institute for Statistical, Social, and Economic Research at the Universityof Ghana, which provided a superb research environment.


982 journal of political economy

the returns to that investment and hence on the nature of propertyrights. In recent years, economists have paid increasing attention to thishypothesis (and brought the argument into the broader public sphere;e.g., De Soto 2000). Economic historians have provided a great deal ofthe evidence that bears on this hypothesis (North 1981; Jones 1986;Mokyr 2002; Engerman and Sokoloff 2003). Additional evidence hasbeen contributed from cross-country regressions of economic growthon a variety of measures of institutional quality (Hall and Jones 1999;Acemoglu, Johnson, and Robinson 2001, 2002; Easterly and Levine 2003;Pande and Udry 2006). This paper joins a growing microeconomicliterature that explores the pathways though which particular institu-tions influence investment or productivity (Place and Hazell 1993; Besley1995; Banerjee, Gertler, and Ghatak 2002; Brasselle, Gaspart, and Plat-teau 2002; Jacoby, Rozelle, and Li 2002; Johnson, McMillan, and Wood-ruff 2002; Gine 2005; Galiani and Schargrodsky 2006; Field 2007; Ra-vallion and Van de Walle 2008). Our aim is to examine one particularmechanism through which the nature of the system of property rightsin a society can shape its pattern of economic activity. We examine theconnection from a set of complex and explicitly negotiable propertyrights over land to agricultural investment and, in turn, to agriculturalproductivity.

There are several potential mechanisms through which propertyrights over land might influence investment in agriculture. Adam Smithfocused attention on the possibility that cultivators’ fears of expropri-ation or loss of control over land on which investments have been mademight deter such investment.1 In addition, access to credit might behindered if property rights are not sufficiently well defined for land toserve as collateral for loans; and an inability to capture potential gainsfrom trade in improved land might reduce investment incentives. Eachof these mechanisms has received a good deal of attention in what hasbecome an important literature. With few exceptions, however, thesepapers “fail to find strong evidence of significant effects of propertyrights on investment” (Besley 1998, 361).

In much of Africa, explicit land transactions—sales, cash rentals, and

1 In his discussion of the Act of Ejectment, which provided for compensation for pastinvestments when a tenant was evicted, Smith writes “when such farmers have a lease fora term of years, they may sometimes find it for their interest to lay out part of their capitalin the further improvement of the farm; because they may sometimes expect to recoverit, with a large profit, before the expiration of the lease. The possession even of suchfarmers, however, was long extremely precarious, and still is so in many parts of Europe.They could before the expiration of their term be legally ousted of their lease. . . . [Butsince the act, in England] the security of the tenant is equal to that of the proprietor.. . . Those laws and customs so favourable to the yeomanry have perhaps contributedmore to the present grandeur of England than all their boasted regulations of commercetaken together” (Smith 1974, vol. 1, bk. III, chap. 2).


profits of power 983

sharecropping—have become more common over recent decades. How-ever, the consensus of the literature is that “the commercialisation ofland transactions has not led to the consolidation of land rights intoforms of exclusive individual or corporate control comparable to West-ern notions of private property” (Berry 1993, 104). Instead, land “issubject to multiple, overlapping claims and ongoing debate over theseclaims’ legitimacy and their implications for land use and the distri-bution of revenue” (Berry 2001, xxi).

The security of farmers’ claims over land is important. In an envi-ronment in which fertilizer is expensive, land is relatively abundant, andcrop returns are sufficiently low, fallowing is the primary mechanism bywhich farmers increase their yields. A significant portion of the agri-cultural land in West Africa is farmed under shifting cultivation, sofallowing remains the most important investment in land productivity.We show that farmers who lack local political power are not confidentof maintaining their land rights over a long fallow. As a consequence,they fallow their land for much shorter durations than would be tech-nically optimal, at the cost of a large proportion of their potential farmoutput.

We provide a brief description of land tenure in southern Ghana inSection II. The primary source of land for farming is the allocation toindividuals of land controlled by that individual’s extended matriline-age, or abusua. The agronomics of intercropped maize and cassava,which is the main farming system in the area, is discussed in SectionII. In that section, we also describe the data and the survey from whichthey are drawn. The most important investment that farmers make intheir land in the study area is fallowing, so we provide a simple modelof efficient fallowing decisions to guide the empirical work in SectionIII.

In Section IV.A, we show that profits per hectare on maize-cassavafarms vary widely across apparently similar plots cultivated by differentindividuals in the same household and that this variation can be attrib-uted to variation in the length of time that these plots have been leftfallow. The essence of our econometric strategy is to examine the effectof an individual’s position in local political and social hierarchies onhis or her fallowing choices on a plot, conditional on plot characteristicsand household fixed effects. In turn, we estimate the productivity effectsof (endogenous) fallowing choices, using the individual’s political andsocial position as instruments for the fallowing choice. Our motivationfor examining the relationship between fallowing decisions and thepolitical and social position of the cultivator is provided by our reviewof the literature on land tenure in West Africa. The exclusion restrictionsare valid within an efficient household because these variables cannotinfluence the within-household shadow prices of inputs or outputs.



However, there are potential unobserved variables that are correlatedwith both productivity and an individual’s social and political status.Therefore, in Section IV.B, we examine in depth—and reject—the pos-sibility that within-household variations in fallowing choices and pro-ductivity are associated with intrahousehold variations in wealth or bar-gaining power. In Section IV.C, we show that individuals with powerfulpositions in local political hierarchies leave their plots fallow for yearslonger than other individuals do, and this effect is stronger on plotsallocated through the prevalent matrilineage allocation process than onplots obtained commercially. In this subsection we also disaggregateofficeholding status into inherited versus noninherited offices to ex-amine the hypothesis of reverse causality running from farming choicesto officeholding. Perhaps most important, we also show that fallow du-rations vary across the different plots cultivated by a single farmer, de-pending on the provenance of the land. Individuals with local politicalpower fallow land that they obtained through the political process ofmatrilineage land allocation significantly longer than they do land ob-tained through other means. This permits us to distinguish betweendeterminants of investment that operate at the individual level (such asunobserved ability) and those that operate at the plot#cultivator in-teraction, such as tenure security.

In Section IV.D, we estimate a model of the annual risk of losing plotswhile they are fallow as a function of individuals’ positions within localpolitical hierarchies and the provenance of the plot. We show that thoseplots that are fallowed for longer durations are exactly the plots thatare more securely held. In Section IV.E, we provide rough estimates ofthe productivity cost of this tenure insecurity and also derive boundsfor discount rates that rationalize the chosen fallow durations, given theestimated productivity of fallowing and the hazard of losing plots whilefallow. Section V presents conclusions.

II. Land Rights and the Farming System in Akwapim

The complexity and flexibility of property rights in West Africa areapparent in our study area in Akwapim, Ghana. Most of the land cul-tivated by farmers in these villages is under the ultimate control of aparamount chief and is allocated locally through the matrilineage (abu-sua) leadership.2 Each farmer in the area cultivates on average fourseparate plots. Land is allocated to individuals for use on the basis oftheir political influence and perceived need.

There is a rich literature that describes the land tenure systems of

2 This is not to say that other forms of ownership/contracts over land do not exist. Wediscuss these less prevalent forms of tenure later.



southern Ghana. The most general principle is that land is “owned” bythe paramount chieftaincy (known as the stool) and is controlled by aparticular abusua subject to that abusua’s members meeting their con-tinuing obligations as subjects of the stool. Individuals, in turn, haverights to the use of farm land by virtue of membership in an abusua.3

This general principle does not define which individual member ofa matrilineage will cultivate which particular plots. Individual claimsover land overlap. Who ends up farming a specific plot is the outcomeof a complex, sometimes contentious, process of negotiation. Moreover,land rights are multifaceted. The act of cultivating a given plot may—or may not—also be associated with the right to the produce of treeson the land, the right to lend the plot to a family member, the rightto rent out the land, the right to make improvements, or the right topass cultivation rights to one’s heirs. A person’s right to establish andmaintain cultivation on a particular piece of land and the extent of herclaims along the many dimensions of land tenure are ambiguous andnegotiable. The situation is further complicated by the tension betweenmatriliny and patriliny as fathers attempt to transfer land rights to theirown children, outside inheritance norms (McCaskie 1995, 77, 277–78;Austin 2004, 174). As a consequence, “people’s ability to exercise claimsto land remains closely linked to membership in social networks andparticipation in both formal and informal political processes” (Berry1993, 104). To summarize, while “in principle, any individual is entitledto use some portion of his or her family’s land, . . . people’s abilitiesto exercise such claims vary a good deal in practice and are often subjectto dispute. Disputed claims may turn on conflicting accounts not onlyof individuals’ histories of land use, field boundaries, or contributionsto land improvements but also their status within the family, or eventheir claims of family membership itself” (Berry 2001, 145).4

In our sample, there are a number of individuals (about 18 percentof the sample) who hold an office of social or political power in their

3 There are numerous descriptions of this principle. See Rattray (1923, 224–41), Klingel-hofer (1972, 132), Wilks (1993), Amanor (2001, 64–76), Berry (2001, 146–56), and Austin(2004, 100).

4 This general pattern of negotiated access to land through membership in a corporategroup is found elsewhere in Ghana, in many parts of West Africa, and in some other areasof Africa, although there is considerable variation in the details. Some examples can befound in Biebuyck (1963), Benneh (1988), Bromley (1989), Bassett (1993), Bruce andMigot-Adholla (1994), Peters (1994), Binswanger, Deininger, and Feder (1995), Fred-Mensah (1996), Sawadogo and Stramm (2000), and Amanor (2001). Summarizing theconclusions of several studies from across the continent, Bassett and Crummey (1993, 20)state that “the process of acquiring and defending rights in land is inherently a politicalprocess based on power relations among members of the social group. That is, membershipin the social group, is, by itself, not a sufficient condition for gaining and maintainingaccess to land. A person’s status . . . can and often does determine his or her capacityto engage in tenure building.”



TABLE 1Perceptions of Land Rights

Percent of Cultivated Plots on WhichRespondent Claims Right to Plots

FallowedMore than6 Years (%)

(5)

DetermineInheritance

(1)Rent Out

(2)Lend Out

(3)Sell(4)

Nonofficeholders 4 15 21 10 18Officeholders 18 37 42 22 26t-test for equality 6.39 6.51 5.56 4.36 2.23Observations 846 847 847 846 813

village or matrilineage. Typical offices include lineage head (abusua-panyin), chief’s spokesman (okyeame), lineage elder, or subchief. Theseare not formal government positions. They instead represent positionsof importance within local political hierarchies. In accordance with theconclusions of other observers, we find in table 1 that such individualsare far more confident than typical farmers of their rights over theircultivated land. Of course, these are their own claims about their rightsalong a limited number of dimensions; below, we examine the rela-tionships between such political power and output and investment de-cisions on these plots and the actual hazard of losing plots while theyare fallow.

A cultivator’s rights over her growing crops, however, are quite secure.Wilks summarizes the principle as “afuo mu y! de!, asase y! ohene de!”(“the cultivated farm is my property, the land is the stool’s”; 1993, 99).5

Plots are virtually never lost while under cultivation. The impact of theparticular form of tenure insecurity that exists in Ghana on certain typesof investment, especially tree planting, therefore, might be quite min-imal.6

However, in the farming system we consider, the most important in-vestment in land quality is leaving land fallow in order to permit soilfertility to regenerate. It is during this period of fallow that one’s rightsover a plot can be lost (see Firman-Sellers 1996, 65; Austin 2004, 333–46). “Because of tenure insecurity under traditional land tenure insti-tutions, there is no strong guarantee that the cultivator can keep fallowland for his or her own use in the future” (Otsuka et al. 2003, 78).Accordingly, we investigate the possibility that the chance that landmight be lost while fallowed leads farmers to reduce the duration of

5 This principle is also supported in the formal court system: “Since colonial times, thecourts have held that while allodial rights to land belong to the stool, families’ rights ofusufruct are secure from arbitrary intervention” (Berry 2001, 145; citing Ollenu [1962]).

6 See Austin (2004) and Pande and Udry (2006) for discussions of the interactionsbetween this land tenure system and the twentieth-century cocoa boom in Ghana.



the fallow period. It is the nexus of a particular form of investment andthese complex and negotiable land rights that has dramatic conse-quences for the overall efficiency of the farming system.

We restrict attention in this paper to the main food crop farmingsystem in the study area, which is an intercropped mixture of maizeand cassava. Approximately three-quarters of the plots cultivated in ourstudy area are planted with these crops. This mixture became the focusof agriculture in the Akwapim region by the 1950s, after swollen shootdisease devastated cocoa production. In addition to maize and cassava,farmers in these villages also cultivate pineapple for export as a freshfruit and a variety of other, more minor, crops.

Soil fertility in the maize and cassava farming system in southernGhana is managed primarily through fallowing: cultivation is periodi-cally stopped in order for nutrients to be restored and weeds and otherpests to be controlled.7 As a result, this farming system exhibits a par-ticularly regular cycle of fallowing and cultivation. Farms are clearedfrom a bush fallow and the cleared brush is burned. The newly clearedplot is cultivated for a single cycle of cassava and maize—long enoughfor one harvest of cassava and two of maize. The cassava harvest oftencontinues over a period of many months, ending approximately 2 yearsafter the initial clearing of the plot. After the cassava is harvested, theplot is returned for another period of fallowing.8 Of 519 plots in oursample, only 61 have been in cultivation for more than 3 years. In mostcases, cultivation continues on these plots because they are primarilyorchards with tree crops; in a few instances these are small garden plotsunder permanent cultivation near the house. We observed no instanceof chemical fertilizers being used to maintain soil fertility on maize-cassava plots. People are aware of fertilizer and use it frequently on thepineapple farms cultivated by some of these households. The absenceof its use on maize-cassava farms indicates to us that fertilizer is lessprofitable than fallowing as a means of maintaining soil fertility in thisfarming system. The fact that no farmer uses fertilizer on maize-cassavaplots, of course, implies that we cannot directly test this conclusion.

Soil scientists working in the area argue that fallow durations of ap-proximately 6–8 years are sufficient to maintain soil fertility in this farm-ing system (de Rouw 1995; Nweke 2004). Ahn (1979, 21) argues that“under forest conditions, both soil organic matter changes and the tran-sition from thicket of young secondary forest re-growth suggest that, inmany areas, a fallow of 6–8 years is a desirable practical minimum: below

7 Amanor (1994, chap. 6) has a useful discussion of fallowing and soil fertility in Krobo,near our study area.

8 This corresponds to the “short fallow” system with one cycle of cultivation describedby Nweke, Spencer, and Lynam (2002). This is the dominant system for cassava cultivationin Africa.



this the soil will be maintained by successive fallows at a lower organicmatter level and level of productivity.”

The median duration of fallow in the plots in our sample is 4 years;the 90th percentile of fallow durations is 6 years. To anticipate resultsthat follow, column 5 in table 1 shows that plots cultivated by individualswho hold local offices are more likely to have been fallowed for morethan 6 years than plots cultivated by others.

To examine this differential and the attendant productivity effects,we use data from a 2-year rural survey in the Akwapim South Districtof the Eastern Region of Ghana.9 Our sample consists of four villageclusters (comprising five villages and two hamlets) with a variety ofcropping patterns and market integration. Within each village clusterwe selected 60 married couples for our sample. Each head and spousewas interviewed 15 times during the course of the 2 years. Every interviewwas carried out in private, usually by an enumerator of the same gender.

In southern Ghana, as in many African societies, agricultural pro-duction is carried out on multiple plots managed separately by individ-uals in households, so each plot in our sample can be identified witha particular individual who controls that plot. The survey was centeredaround a core group of agricultural activity questionnaires (plot-levelinputs, harvests, sales, credit) that were administered during each visit.The purpose of this high frequency was to minimize recall error onreports of plot-level inputs and outputs. In addition, about 35 othermodules were administered on a rotating basis. We also administeredan in-depth plot rights and history questionnaire and mapped each plotusing a global positioning system. We supplemented this with data onsoil fertility: the organic matter and pH of approximately 80 percent ofthe plots was tested each year. We also make use of data on educationand individual wealth. It is possible to collect the latter because of thequite separate accounts that are kept by husbands and wives.

Table 2 reports summary statistics on the variables we use in thispaper’s analysis. Plot profits are calculated with household labor valuedat gender-village-survey round specific median wages. Given that we areexamining the role of political power in tenure security, we have sep-arated the summary statistics by the officeholding status of the individ-ual. Average per-hectare profits and yields are comparable on the plotscultivated by office holders and nonofficeholders, but officeholders cul-tivate larger plots. Inputs and measured soil organic matter and pH ofplots are similar across officeholding status. The average duration ofthe last fallow period is almost a year longer for officeholders, andofficeholders have had control over their plots for much longer thannonofficeholders. Officeholders are significantly more likely to be cul-

9 The data and documentation are available at http://www.econ.yale.edu/cru2.



TABLE 2Summary Statistics

Variable

Officeholders Nonofficeholders

FtFMeanStandardDeviation Mean

StandardDeviation

Plot-level data:Profit#1,000 cedis/hectare 649.10 2,374.87 580.59 6,864.34 .11Yield#1,000 cedis/hectare 1,490.32 2,850.93 1,615.14 7,353.06 .18Hectares .48 .62 .31 .30 4.26Labor cost#1,000 cedis/hectare 651.39 1,155.59 883.14 2,223.01 1.11Seed cost#1,000 cedis/hectare 282.12 612.24 243.08 719.98 .45pH 6.36 .71 6.34 .75 .22Organic matter 3.22 1.06 3.13 1.08 .67Last fallow duration (years) 4.83 4.23 3.93 2.65 2.60Length of tenure (years) 16.13 16.10 7.32 9.47 7.26Plot same abusua as individual .66 .47 .56 .50 1.79Plot obtained via commercial

transaction .25 .43 .30 .46 1.17Observations 122 484

Individual-level data:Gender (1 p female) .11 .32 .40 .49 3.73Age 51.92 13.47 40.08 12.21 5.41Average assets#1,000 cedis 1,475.52 1,767.18 620.39 902.57 4.71Years of schooling 7.56 6.98 7.09 4.92 .501 if mother was a trader .09 .29 .24 .43 2.231 if mother was a farmer .89 .32 .72 .45 2.351 if father was a farmer .82 .39 .79 .41 .461 if father was an artisan .07 .25 .11 .31 .761 if father was a civil servant .09 .29 .09 .29 .021 if father was a laborer .00 .00 .00 .07 .461 if first in village of family .11 .32 .23 .42 1.82Years family or respondent has

been in village 64.80 41.63 53.50 39.44 1.72Number of wives of father 2.82 1.71 2.14 1.20 3.18Number of children of father 12.04 7.32 10.84 6.51 1.10Parity of mother in father’s wives 1.71 1.47 1.30 .64 2.941 if fostered as a child .58 .50 .69 .46 1.46Size of inherited land .62 .83 .13 .39 6.101 if mother had any school .04 .21 .12 .32 1.431 if father had any school .16 .37 .31 .46 2.09Observations 45 207

tivating plots that come from their own matrilineage than others. Thereis some indication that officeholders cultivate fewer plots obtainedthrough commercial transactions. Approximately half of these com-mercial transactions are sharecropping contracts, and half are based onfixed rent.

Officeholders are much more likely to be male and are older, richer,and better educated than other farmers in their villages. Their motherswere more likely to be farmers, and their fathers had more wives. Theyare less likely to be the first of their family to settle in the village, andtheir families have lived in the village longer. They claim to have in-



herited more land (although we are skeptical about the accuracy of thisparticular variable because we were not able to physically measure thearea claimed to be inherited, and farmer estimates of the areas of theplots they do cultivate were extremely inaccurate [Goldstein and Udry1999]). The parents of officeholders were less likely to have been ed-ucated than others in the village, perhaps reflecting the age of theofficeholders.

III. Productivity in a Fallow Farming System

An individual’s decisions regarding the optimal time path of fertilityand of agricultural output from a given plot in such a system depend,among other things, on the opportunity cost of capital to that individualand his or her confidence in the ability to reestablish cultivation on theplot after fallowing.

Consider an individual i (in household h) with control over a set Pi

of plots of land (indexed by p). We assume that i’s aim is to managefertility to maximize the present value of the stream of profits she canclaim from this land.10 The salient decision facing this individual is thelength of time she should leave each plot fallow before cultivation.Considered in a stationary environment, this corresponds precisely tothe optimal harvest problem solved long ago by Faustmann ([1849]1995).

Suppose that the profit (per hectare) that can be generated fromcultivating a plot depends on the time that the plot has been left fallowaccording to the strictly concave and increasing function , wherep (t)p

t denotes the number of years the plot has been left fallow. Denote bythe household-specific annual discount rate. Let the (constant) like-rh

lihood of losing plot p during a year in which it is fallow be . Theqp

discussion in Section II implies that may vary according to i’s statusqp

in local political hierarchies and according to the manner in which iacquired plot p.

10 In general, of course, this assumption is consistent with utility maximization only iffactor and insurance markets are complete (Krishna 1964; Singh, Squire, and Strauss1986). However, we will focus on comparisons across plots within households and alsoacross different plots cultivated by the same individual. If households are Pareto efficient(as in Chiappori [1988]), then by the second welfare theorem there exist (household-specific) shadow prices such that fertility management decisions correspond to those thatmaximize the present discounted value of the stream of profits at those shadow prices.Similarly, when we examine fertility decisions across plots of a particular individual, wewill be assuming that the allocation of resources across plots cultivated by that singleindividual is Pareto efficient. In this case, there are individual-specific shadow prices suchthat the present discounted value of the stream of profits from each of the individual’splots is maximized.



Suppose for the moment that cultivation itself takes no time. Thenthe expected present discounted value of profits from i’s plots is

! nt tp p1 " q [(1 " q )/(1 # r )]p p hp (t ) p p (t ) . (1)! ! !p p p p( ) tp1 # r 1 " [(1 " q )/(1 # r )]p!P np1 p!Pi ih p h

The individual maximizes (1) with respect to . Let denote the′t p (t)p p

first derivative of the profit function. The concavity of ensures ap (7)p

unique optimal fallow duration for each plot ( ), which is definedt*pimplicitly by

′p (t*) ln [(1 " q )/(1 # r )]p p p pp " . (2)t*pp (t*) 1 " [(1 " q )/(1 # r )]p p p h

The optimal fallow duration falls with increases in the likelihood thatthe individual will lose the plot and with the discount rate. It is apparentfrom (2) that for any two plots p and q cultivated by the same individual,if they are similarly securely held ( ) and have similar physicalq p qp q

characteristics ( ), then the optimal fallow durations are thep (t) { p (t)p q

same on each plot ( ). The same holds for any two plots withint* p t*p q

a given household if the household is Pareto efficient.11

We supposed “for the moment” that cultivation occurred instanta-neously. In fact, as we discussed above, the cultivation cycle in this farm-

11 This general message is robust to imperfect markets that provide an incentive forindividuals to adjust harvest periods to smooth factor demand. Consider, e.g., a simplediscrete-time model and a household with two plots. In any given year, if the householdcultivates one plot that had been fallowed the previous year, it earns Y. If it cultivates bothplots, each having been fallowed, it earns only from each plot, . This reflects thevY v ! 1costs of extending cultivation beyond the single plot where labor or other input marketsare imperfect. If a plot was not left fallow the previous year, it yields , (so fallowingdY d ! 1/2is potentially productive). If two unfallowed plots are cultivated, total return is ; if2dvYone unfallowed and one fallowed plot are cultivated, the return is . Let the statevY(1 # d)variable denote the number of plots fallowed last period. The discount factors ! (0, 1, 2)is b, and the household is risk neutral and maximizes the discounted stream of futurereturns. The value functions arevs

v p max {bv , Y # bv , 2vY # bv },2 2 1 0

v p max {bv , Y # bv , vY(1 # d) # bv },1 2 1 0

v p max {bv , dY # bv , 2dvY # bv }.0 2 1 0

The choices in each maximand correspond to fallowing two, one, or zero plots, respectively.Depending on parameter values, there are a number of possible equilibria. The interestingcase is the two equilibria

0 r 2, 1 r 1, 2 r 1

0 r 0, 1 r 1, 2 r 1.In the first pattern, the equilibrium of cultivating one of the two plots each year rapidlyemerges, regardless of the initial state of fallowing. However, for sufficiently low b, thesecond pattern emerges. If the cultivator starts with none of her plots fallowed ( ),s p 0she is sufficiently impatient that she does not ever begin fallowing. The key point is thatin the steady state of any equilibrium, each similar plot is treated identically.



ing system occurs over a period of 2 years. During the period of culti-vation, there is no chance that the plot will be lost. This fact does notchange the essence of this argument. Accounting for the 2-year periodduring which cultivation occurs changes the expected present dis-counted value of profits of the plots cultivated by i to

t "2 tp p(1 " q ) /(1 # r )p hp (t ) . (1′)! p p t "2 tp p1 " [(1 " q ) /(1 # r ) ]p!Pi p h

Equation (2) becomes considerably more complex without changingthe comparative static conclusions at all.

Given imperfect financial and labor markets in rural Ghana, it isunlikely that the opportunity costs of capital or labor are identical acrossplots cultivated by individuals in different households. However, theywill be the same across plots cultivated by the same individual, and ifhouseholds allocate resources efficiently across household members,then they will be identical across plots within households. These ob-servations form the basis of our initial empirical work.

We begin by supposing that households allocate resources efficiently.If so, the marginal value products of inputs used on farm operationsare equated across plots within households. We do not assume that inputcosts or the opportunity cost of capital is similar across households.Within the household, plots of similar fertility should be cultivated sim-ilarly. Moreover, we have seen in (2) that the optimal fallowing perioddoes not vary across plots within the household, except as a functionof their physical characteristics or of the security with which they areheld.

So we can define profits on plot p cultivated by individual i in house-hold h at time t as a function only of the characteristics of that plot:

p(t*(X , q ), X ), (3)p p p p

where is defined as a vector of fixed characteristics of plot p andXp

is the duration of the last fallow on plot p. A first-order approximationt*pof the difference in profits across plots within a household is

"p "p¯ ¯p(t*, X ) " p(t , X ) ≈ (t* " t ) # (X " X ), (4)p p h p h ph hp pp p"t "X

where is the household in which the cultivator of plot p resides, andhp

bars indicate averages of characteristics over the plots cultivated byhousehold .hp

We rewrite (4) as

p p at* # X b # gG # l # ! , (5)pt p p p h ,t ptp

where is the profit measured on plot p in year t, a is , b isp "p/"tpt



, and is the gender of the individual who cultivates plot p. The"p/"X Gp

term is a fixed effect for the household-year, and is an error terml !h ,t ptp

(that might be heteroskedastic and correlated within household-yeargroups) that summarizes the effects of unobserved variation in plotquality and plot-specific production shocks on profits. An exclusionrestriction of the model is that . In an efficient household, theg p 0identity of the cultivator is irrelevant for profits.

Within the vector we include a variety of plot characteristics: size,Xp

toposequence, direct measures of soil quality (the soil pH and organicmatter content), and the respondent-reported soil type classified intoclay, sandy, or loam. These soil types might affect profits and inputsthrough their different nutrient and moisture retention capacities,among other factors.

Equation (2) implies that is chosen optimally. We can expectt* t*p p

to be correlated with , even conditional on , because it may re-! lpt h ,tp

spond to the same unobserved attributes of the plot that influenceprofits. From (3), we see that the appropriate instrument for is —t* qp p

the security of tenure over that plot. However, is unobserved. There-qp

fore, on the basis of the discussion of Section II, we collected a set ofvariables that represent the cultivator’s position in local social and po-litical hierarchies. These variables might influence her tenure securityand thus her choice of optimal fallow duration, and we estimate (5)using these as instrumental variables.

IV. Results

A. Fallowing and Within-Household Productivity Variation

We begin with what we expect is the counterfactual assumption thatthere is complete tenure security on all plots in our sample, whichimplies that for all plots. In this case, equation (2) implies thatq p 0p

optimal fallow duration is a function only of and household-t* Xp p

specific shadow prices. Equation (3) becomes , wherep(t(X ), X ) t(7)t p p

is defined implicitly by (2) evaluated at . Within-household differencesrh

in plot profits (4) depend only on differences in plot characteristics, sowe modify (5) and estimate

˜ ˜ ˜˜p p X b # gG # l # ! . (6)pt p p h ,t ptp

In (6) isb

"p "t "p# ;

"t "X "X

that is, it captures both the direct and the indirect (through fallowingchoice) effects on plot profits of variation in plot characteristics. The



TABLE 3Profits and Gender

Dependent Variable: Profit#1,000 Cedis/Hectare

OLS(1)

OLS(2)

OLS(3)

Gender: 1 p woman"913(365)

"985(468)

"1,683(380)

Plot size decile:

2198

(486)1,049(571)

1,646(265)

3689

(507)1,239(590)

749(265)

4655

(508)1,806(591)

1,557(364)

525

(502)883

(583)923

(147)

6377

(489)1,447(581)

819(222)

7"79(494)

1,206(548)

628(252)

8"389(520)

593(594)

"180(259)

946

(513)705

(633)420

(261)

10"383(597)

"17(693)

"693(338)

Soil type:

Loam629

(342)35

(396)"21(151)

Clay226

(381)"58(463)

122(321)

Toposequence:

Midslope"364

(1,110)339

(1,581)"705(493)

Bottom"45

(1,104)661

(1,569)"722(552)

Steep"800

(1,153)"83

(1,610)476

(695)

pH"122(247)

"202(78)

Organic matter"26(150)

135(49)

Observations 888 614 575Fixed effects Household

#yearHousehold

#yearSpatial (250meters) and

household#yearNote.—Standard errors are in parentheses.

exclusion restriction remains in force, under the joint null hy-g p 0pothesis that the household is Pareto efficient and that there is novariation in tenure security across plots.

We present estimates of (6) in table 3. Recall that the results areinterpreted in terms of deviations from household-year means for cas-sava-maize plots. We do not expect returns to be equalized across house-



holds or years because of imperfect factor markets within villages. Col-umn 1 presents ordinary least squares (OLS) results.12 The most strikingresult concerns gender: women achieve much lower profits than theirhusbands. Conditional on household-year fixed effects and on the ob-served characteristics of their plots, women get 900,000 cedis less inprofits per hectare then their husbands. Average profits per hectare areapproximately 600,000 cedis, so this is a very large effect. Given dimin-ishing returns, a systematic difference in the cassava/maize profits onsimilar plots of men and women within a household in a given yearrejects our joint null of Pareto efficiency within households and theassumption that tenure security is the same across plots within a house-hold. The literature contains similar results in some other West Africancontexts (Udry 1996; Akresh 2005); those papers have interpreted it asa violation of the null hypothesis of within-household Pareto efficiency.Here, we raise the possibility that the within-household dispersion inyields on similar plots may arise from the land tenure system.

Another possible explanation for the gender differential in farm prof-itability is that women farm plots that are of lower exogenous qualitythan those of their husbands. In column 2, we add additional infor-mation on soil quality, in the form of data on the soil pH and organicmatter content measured on most plots.13 Differences in this dimensionof measured soil quality do not help explain the gap in profits betweenhusbands and wives.

It is possible that the plots of husbands and wives are physically sys-tematically different from each other along dimensions that we do notobserve. The different profitability of their plots might be a consequenceof these unobserved differences in fundamental plot characteristics.These unobserved differences in physical characteristics might have todo, for example, with variations in soil physical structure or quality thatare finer than we observe or with differences in moisture or patternsof water runoff. In the Akwapim region, these relatively fine physicalcharacteristics of land tend to vary gradually over space. Plots close toeach other (within a few hundred meters) are more likely to be verysimilar than plots separated by larger distances. This can be seen in

12 The standard errors in all our specifications use limiting results for cross-sectionestimation with spatial dependence characterized by physical distance between plots. Spa-tial standard errors are calculated using the estimator in Conley (1999) with a weightingfunction that is the product of one kernel in each dimension (north-south, east-west). Ineach dimension, the kernel starts at one and decreases linearly until it is zero at a distanceof 1.5 kilometers and remains at zero for larger distances. This estimator allows generalcorrelation patterns up to the cutoff distances.

13 We lose some plots because of the administrative difficulties of conducting such alarge number of soil tests. In addition, soil pH and organic matter content are likely torespond to fallowing decisions; hence, in most of the results that follow these variablesare excluded.



figure 1, which is a map of the plots in one of the villages. This mapalso shows houses (as stars) and paths. The other villages are organizedsimilarly.

Therefore, we generalize (6) to permit a local neighborhood effectin unobserved land quality that could be correlated with gender andthe other regressors. With some abuse of notation, let denote bothNp

the set of plots within a critical distance of plot p and the number ofsuch plots. We construct a within estimator by differencing away thesespatial fixed effects:

1 1 1˜ ˜p " p p X " X b # g G " G! ! !pt qt p q p q( ) ( )N N Nq!N q!N q!Np p pp p p

1 1˜ ˜ ˜ ˜# l " l # ! " ! . (7)! !h t h t pt qtp qN Nq!N q!Np pp p

In column 3 of table 3 we define the geographical neighborhood ofeach plot using a critical distance of 250 meters. If the component ofunobserved land quality that is correlated with the regressors in (7) isfixed within this small neighborhood, then the spatial fixed-effect es-timator removes this potential source of bias. Wives achieve much lowerprofits than their husbands, even on plots that are within 250 metersof each other.

Husbands and wives achieve very different profits on plots that sharevery similar fundamental characteristics. However, these estimates ne-glect the anthropogenic differences in soil fertility that emerge as aresult of the varying fallowing histories of their plots. If tenure securityis not the same on all plots and this variation is correlated with gender,then fallowing choices might systematically vary across plots that oth-erwise look similar. Hence, in table 4 we present estimates of equation(5).

In column 1 we present OLS estimates that ignore the potential en-dogeneity of t*. Unsurprisingly, given the discussion of soil fertility inSection II, we find that longer fallow durations are strongly associatedwith higher profits. Perhaps more important, the coefficient on genderfalls by more than half and is statistically insignificant. Conditional onfallow duration, we can no longer reject the hypothesis that profits aresimilar on men’s and women’s plots within a household-year.

The optimal duration of fallowing on a plot depends on unobservedplot and individual characteristics and so is treated as endogenous incolumns 3, 4, and 6. We use a set of variables based on the social andpolitical family background of the cultivator as instruments for the du-ration of the most recent fallow. In Section II, we saw that an individual’ssecurity of tenure on a given plot is influenced by his or her positionin local social and political hierarchies. Conditional on the assumption


Fig.

1.—

Plot

sof

sam

ple

mem

bers

inon

evi

llage


998

TAB

LE

4Pr

ofits

and

Fall

owD

urat

ion

OL

SO

LS

IVIV

OL

SIV

DV

:Pro

fit#

1,00

0C

edis

/Hec

tare

(1)

DV

:Fal

low

Dur

atio

n(2

)

DV

:Pro

fit#

1,00

0C

edis

/Hec

tare

(3)

DV

:Pro

fit#

1,00

0C

edis

/Hec

tare

(4)

DV

:Fal

low

Dur

atio

n(5

)

DV

:Pro

fit#

1,00

0C

edis

/Hec

tare

(6)

Fallo

wdu

ratio

n(y

ears

)*14

5(4

8)54

1(2

33)

662

(261

)32

6(5

8)G

ende

r:1p

wom

an"

473

(393

)"

.58

(.67

)13

0(5

55)

"68

8(7

32)

".3

4(.

24)

328

(148

)A

ge"

118

(62)

16

year

sof

scho

ol"

286

(720

)1

iffir

stof

fam

ilyin

tow

n"

.44

(.66

).3

0(.

19)

Year

sfa

mily

/res

pond

ent

lived

invi

llage

".0

1(.

01)

.01

(.00

)


999

1if

resp

onde

ntho

lds

trad

ition

alof

fice

3.91

(1.1

1)1.

82(.

33)

Num

ber

ofw

ives

offa

ther

.39

(.35

).5

0(.

13)

Num

ber

offa

ther

’sch

ildre

n"

.08

(.07

)"

.01

(.02

)Pa

rity

ofm

othe

rin

fath

er’s

wiv

es"

.44

(.41

)"

.43

(.32

)1

iffo

ster

edas

ach

ild.8

6(.

74)

.37

(.34

)Si

zeof

inhe

rite

dla

nd"

.29

(.63

)"

.45

(.21

)1

ifm

othe

rha

dan

yed

ucat

ion

".8

7(1

.17)

1.18

(.61

)1

iffa

ther

had

any

educ

atio

n"

.13

(.80

)"

1.16

(.31

)O

bser

vatio

ns76

075

575

567

260

960

9Fi

xed

effe

cts

Hou

seho

ld#

year

Spat

ial(

250

met

ers)

and

hous

ehol

d#ye

arJ-s

tatis

ticof

over

iden

tifyi

ngre

stri

ctio

nsx

2 (16)

p10

.25

x2 (9

)p

2.25

x2 (9

)p

2.43

F-te

stof

inst

rum

ents

F(10

,415

)p

2.10

F(10

,272

)p

20.8

3N

ote.

—A

llre

gres

sion

sin

clud

eth

epl

otch

arac

teri

stic

sus

edin

tabl

e3,

col.

1.St

anda

rder

rors

(in

pare

nthe

ses)

are

cons

iste

ntfo

rar

bitr

ary

spat

ial

corr

elat

ion.

*Tr

eate

das

endo

geno

usin

cols

.3,4

,and

6.In

stru

men

tsar

eas

indi

cate

din

cols

.2an

d5.



of Pareto efficiency within the household, these variables cannot influ-ence the shadow price of factors of production or output and hencedo not enter (3) except via , the security of tenure. We test the over-qp

identifying restrictions implied by this assumption in table 4 and relaxthe household efficiency assumption by moving to a within-individualprocedure in Section IV.C.

The instrument set includes the indicator that the individual holdsan office of local social or political power as in table 2. In addition, weinclude more subtle dimensions of the individual’s status within thevillage and matrilineage. These include two indicators of the length oftime the cultivator’s household has been resident in the village. Newermigrants to the village have a shorter history of local land use, and wesaw in Section II that the history of land allocations can play a role inthe security with which an individual holds a plot. We also include thenumber of wives of the individual’s father and the parity of the indi-vidual’s mother in that set of wives. In a polygamous union, the positionof a wife in the order of marriages is important for her children’s claimsover property, among other things. We also include the number ofchildren of the individual’s father and an indicator of whether the in-dividual was fostered as a child. Each of these variables is an attemptto capture an aspect of the individual’s place within his or her matri-lineage, which the literature implies would influence . Finally, weqp

include measures of the education of the individual’s parents as themost important indicator of the parent’s social status.

The results of the first-stage regression are presented in column 2 oftable 4. The instruments are jointly significant determinants of the du-ration of fallow on a plot. Officeholders fallow their plots much longerthan others. We interpret this first-stage regression as preliminary evi-dence that the local social and political status of individuals does influ-ence their security of tenure and that this in turn permits them to leaveplots fallow for longer periods of time. This hypothesis is examined inmore detail below in Sections IV.B–D.

The entire difference between profits on husbands’ and wives’ plotsis attributable to the longer fallow periods on men’s plots. In column3, we show that conditional on fallowing choices, there is no genderdifferential in profits within households. Instead, we find a strong pos-itive correlation between fallow periods and profits: each additional yearof fallowing is associated with about 550,000 cedis of additional profitsper hectare. This is a very large effect, given a standard deviation offallowing of about 3 years. The instrumental variable (IV) estimate ofthe effect of fallow duration on profits is more than thrice that obtainedvia OLS, implying that fallowing is negatively correlated with other un-observed determinants of profitability on plots. Farmers appear to com-pensate for worse plot conditions by extending fallow durations.



Within-household variations in age and education are not driving thevariation across plots in fallow durations or profits. We saw in table 2that officeholders tended to be older than other cultivators. However,in column 4 of table 4, we show that neither age nor education accountsfor any of the difference in profits per hectare on plots that have longerfallow durations.

As before, there is a potential concern with unobserved variation inexogenous plot characteristics. If these unobserved characteristics arecorrelated with the social and political status of the cultivator, then theIV estimator is inconsistent. It is possible, for example, that officeholdersget land that is better than average and that output is higher on thoseplots. If it is also the case that these plots are left fallow for longerperiods (perhaps for reasons orthogonal to productivity), then we couldsee the pattern of results displayed in columns 2 and 3. Therefore, weestimate (5) with spatial fixed effects as well in columns 5 and 6. Thestrong effect of fallow durations on plot-level profits remains apparentconditional on these spatial fixed effects, and we now find that wivesachieve even larger profits than their husbands once we condition onfallow duration.

B. Bargaining, Wealth, and Fallowing Decisions

In this subsection we examine more carefully the determinants of thisvariation in fallowing choices across plots within households. First, weconsider the possibility that inefficient fallowing is a consequence of aninefficient bargaining process within the household. Second, we ex-amine the hypothesis that individuals (within a household) of differentsocial and political status face different opportunity costs of capital andthat these differences induce them to choose different fallow durations.

We see in columns 2 and 5 of table 4 that, within households, indi-viduals fallow longer if they have political office. When we control forspatial effects, the length of fallow is associated with the number of wivesof their father, their land inheritance, and their parents’ education.These indicators of social and political status could be associated withintrahousehold bargaining power. Perhaps the variation in fallowing isa consequence of some inefficient bargaining process within the house-hold (inefficient because an efficient allocation within the householdwould equalize fallow durations across similar plots). We will addressthis possibility in three steps. First, in table 5, we show that the wives ofofficeholders do not have characteristics that distinguish them from thewives of nonofficeholders along the dimensions that determine fallow-ing, aside from having an officeholder as a spouse. Nor are their at-tributes vis-a-vis the wives of nonofficeholders those typically associatedwith diminished intrahousehold bargaining power in the West African



TABLE 5Characteristics of Wives of Officeholders and Nonofficeholders:

Individual-Level Data

Variable

Officeholders Nonofficeholders

FtFMeanStandardDeviation Mean

StandardDeviation

Age 45.52 12.86 36.17 13.01 3.65Average assets#1,000 cedis 720.48 1,202.12 324.46 243.01 3.27Years of schooling 2.85 3.92 5.11 4.31 2.701 if mother was a trader .18 .39 .26 .44 .931 if mother was a farmer .82 .39 .69 .46 1.421 if father was a farmer .76 .44 .79 .41 .451 if father was an artisan .09 .29 .06 .23 .711 if father was a civil servant .15 .36 .13 .34 .301 if father was a laborer .00 .00 .01 .10 .551 if first in village of family .52 .51 .25 .44 2.91Years family or respondent

has been in village 38.36 32.81 48.78 41.43 1.32Number of wives of father 2.09 1.01 2.13 1.10 .19Number of children of

father 8.85 4.95 11.71 7.57 2.04Parity of mother in father’s

wives 1.33 .60 1.30 .74 .241 if fostered as a child .76 .43 .72 .45 .48Size of inherited land .12 .33 .04 .23 1.581 if mother had any school .03 .17 .18 .38 2.161 if father had any school .26 .45 .41 .49 1.50Observations 38 118

context. For example, wives of officeholders are older and wealthierthan wives of other men. Second, the fact that their husbands are of-ficeholders may imply that the wives have relatively low weight in someinefficient intrahousehold bargaining process. Therefore, in SectionIV.C, we will show similar magnitudes of variation in fallowing choicesacross plots cultivated by individuals, where inefficient bargaining doesnot arise as a possible explanation. Third, in Section IV.D, we estimatea hazard model that provides direct evidence of land tenure insecuritythat coincides with these results.

An alternative hypothesis is that officeholders fallow their plots morethan others because they face a lower opportunity cost of capital. It isplausible that (within a household) relatively wealthy individuals are lesscredit constrained and therefore choose longer fallow periods. We areable to measure individual wealth holdings because in West Africa mostnonland assets are held by individuals rather than by households. Wealthin this exercise is defined as the value of individual holdings of financialassets, stocks of agricultural inputs and outputs, stocks of goods fortrading, physical assets and working capital of individual businesses,livestock, farm equipment, and consumer durables.



Of course, individual wealth may be correlated with unobserved char-acteristics of the plots cultivated by the individual. Therefore, we esti-mate the determinants of the duration of the last fallow period treatingcurrent wealth as endogenous, using the occupational background ofthe cultivator’s parents as instruments for wealth. The relevant condi-tioning information includes all the measures of the social and politicalbackground of the cultivator that appeared in table 4, including theamount of inherited land, traditional officeholding status, and migratoryhistory. The identification assumption is that conditional on these otherdimensions of the cultivator’s background, parental occupation influ-ences fallowing decisions only through its effect on wealth. The justi-fication for this assumption is that information about the technical prop-erties of fallowing in this farming system is well distributed given itslong dominance in the region. The estimates in Section IV.C are robustto deviations from this assumption.

The first-stage estimates of the determinants of current wealth arereported in column 1 of table 6. The instruments are jointly highlysignificant determinants of current wealth. Current wealth is much lowerif the cultivator’s mother was a farmer rather than the excluded categoryof trader (or a few other miscellaneous occupations). Current wealthis much higher if the cultivator’s father had an office job and somewhathigher if the cultivator’s father was a farmer, relative to the excludedcategory of laborer/artisan. Several of the conditioning variables arealso strongly related to current wealth: current wealth is strongly posi-tively correlated to the number of wives of the father and to the parityof one’s own mother in that set and negatively related to the numberof children of one’s father. Individuals whose families have recentlymigrated to the village tend to be wealthier, and those who were fosteredas children poorer. As we saw in the summary statistics, officeholderstend to be wealthier than others.

Current wealth is well determined by the occupations of one’s parentsbut in turn has nothing to do with fallowing decisions. In column 2 wepresent the fixed-effect (spatial and household) IV estimates of thedeterminants of fallow duration with current wealth treated as endog-enous. The coefficient on current wealth is quite precisely estimated tobe near zero: the point estimate implies that individuals with 1 millioncedis in additional wealth (mean wealth is 700,000 cedis) reduce thefallow duration on their plots by about a month, and the coefficient isnot significantly different from zero. Moreover, the estimated impact ofofficeholding on fallowing decisions is unchanged from our earlier spa-tial fixed-effect specification. These results provide no support for thehypothesis that variations within the household in the cost of capital lieat the root of variations in fallowing across the plots cultivated by house-hold members.



TABLE 6Fallow, Wealth, and Land Owned

OLS IV OLSDV: Wealth

(#1,000 Cedis)(1)

DV: FallowDuration

(2)

DV: FallowDuration

(3)

Wealth (#1,000 cedis)* ".001(.001)

Gender: 1pwoman 32(107)

".01(.20)

".27(.23)

Area on other plots (hectares) ".16(.07)

1 if first of family in town 145(89)

.06(.32)

.22(.28)

Years family/respondent lived invillage

8(1)

.01(.01)

.01(.00)

1 if respondent holds traditionaloffice

497(174)

2.35(.56)

2.01(.36)

Number of wives of father 128(36)

.46(.18)

.33(.17)

Number of father’s children "46(10)

".05(.04)

.00(.02)

Parity of mother in father’s wives 141(64)

".16(.29)

".40(.34)

1 if fostered as a child "152(86)

.20(.30)

.38(.34)

Size of inherited land "262(118)

".69(.25)

".33(.23)

1 if mother had any education "318(239)

.18(.50)

.67(.52)

1 if father had any education "84(91)

".76(.42)

".83(.41)

1 if mother was a farmer "658(232)

1 if father was a farmer 357(111)

1 if father had an office job 696(168)

Observations 413 413 413Fixed effects Household and spatial fixed effects (250 meters)J-statistic of overidentifying

restrictions x2(2) p 1.40F-test of instruments F(3, 409) p 6.51

Note.—Standard errors are in parentheses.* Treated as endogenous. Instruments are as indicated in col. 2.

In column 3, we examine another dimension of wealth: the total landarea controlled by the individual (minus the area on the plot underconsideration). We find that fallow durations are decreasing in the totalarea controlled by the individual. The standard deviation of area onother plots is approximately 1 hectare; increasing area by that magnitudeis associated with a relatively small but statistically significant decline infallowing of approximately 2 months. This result should be treated with



TABLE 7Fallow Duration and Plot OriginDependent Variable: Fallow Duration

OLS: All Plots

(1) (2) (3) (4)

Gender: 1pwoman ".35(.20)

".36(.20)

".28(.22)

".48(.24)

1 if officeholder 1.73(.49)

.68(.59)

1 if holds inherited office 2.28(.93)

1.49(.65)

1 if noninherited office 1.29(.53)

".52(.95)

Plot in same abusua as cultivator .25(.21)

.36(.27)

Cultivator holds office#Plot insame abusua as cultivator

3.24(.89)

Cultivator holds inherited of-fice#Plot in same abusua ascultivator

1.63(1.57)

Cultivator holds noninherited of-fice#Plot in same abusua ascultivator

2.92(1.01)

Observations 402 402 402 402Fixed effects Household and spatial fixed effects

(250 meters)Note.—Standard errors are in parentheses.

caution because it is plausible that the total area cultivated by an in-dividual is correlated with unobserved variables that influence fallowingchoices. Unfortunately, we cannot construct a theoretical argument forthe existence of variables that influence the area of land cultivated byeach individual that do not also influence that individual’s tenure se-curity and thus fallow duration. However, we can see from these resultsthat the strong effect of officeholding on fallowing durations is not asimple consequence of officeholders having more land and thereforemechanically fallowing land for longer.

C. Political Power, Tenure Security, and Investment in Fertility

The strongest and most consistent of our results is that those who holda local social or political office fallow their land longer than others intheir households and, as a consequence, achieve higher profits. Theremainder of the paper focuses on the relationship between office-holding and investment in land fertility. Column 1 of table 7 providesthe baseline result: conditional on household and spatial fixed effectsand on the same plot characteristics included in table 3, officeholdersleave their plots fallow for almost 2 years longer than others in the



same household. This result is very similar to that shown in column 5of table 4.

We treat officeholding status as exogenous to fallow duration on farm-ers’ plots. It is possible, though, that offices are awarded to individualsin part on the basis of their decisions as farmers. In nearby northernNigeria one common office is sarkin noma, “chief farmer,” which is oftenawarded to a particularly innovative or successful farmer. One mightnot want to treat such an office as exogenous in a regression such asthat reported in column 1 of table 7. As a first step, therefore, we dividethe offices reported in our data into two categories: the first is the setof offices that are typically inherited (e.g., abusuapanyin, or lineagehead). The second are offices that are not inherited (e.g., village youthchief). We estimate the coefficients of these two types of office separatelyin column 2. In both cases, there is a statistically significant positiverelationship between officeholding and fallow durations. The point es-timate is stronger for inherited offices than for noninherited offices,although the difference is not statistically significant. This exercise pro-vides no evidence that the strong positive relationship between office-holding and fallow durations is being driven by a simple reverse causalitybetween farming performance and ascent to office. The more subtleworry that officeholders have unobserved characteristics that might beassociated with longer fallow durations is addressed in the within-indi-vidual analysis that follows in table 8. For the remainder of the paper,we will report results both for the aggregate set of officeholders anddisaggregated by type of office.

All land in our sample can be traced to a specific matrilineage,whether it was allocated through the matrilineage-based political processof land allocation or not. Approximately 60 percent of the plots in oursample are controlled by the matrilineage of the cultivator (table 2).There are several mechanisms through which individuals can come tobe cultivating plots that are not of their own matrilineage. Most com-monly, this occurs as a consequence of a commercial transaction orbecause the land is obtained from one’s spouse or father, who are oftenmembers of a different matrilineage. We hypothesize that holding alocal political office is particularly effective in improving an individual’ssecurity of tenure over those plots that are allocated through the politicalprocess of allocating matrilineage land as described in Section II.

In column 3 of table 7, we present a household and spatial fixed-effects regression of the determinants of fallow duration focusing onthe provenance of the plot and its interactions with the political statusof the cultivator. The estimates show that the fallowing differential thatwe observe between those who hold a local political office and thosewho do not occurs only on land that is allocated through the matrilin-eage. On land obtained commercially or through immediate family,



there is no statistically significant difference between the fallowing be-havior of officeholders and that of other individuals. For nonoffice-holders, there is no statistically significant difference between the fallowdurations on plots that they cultivate that originate in their own ma-trilineage and on those plots obtained from other sources. However, onland allocated by the matrilineage, officeholders have fallow durationsthat are more than 3 years longer than those of nonofficeholders.

A similar pattern emerges in column 4, where we disaggregate be-tween inherited and noninherited offices. Once again, for nonoffice-holders, there is no statistically significant difference in fallowing onland obtained from the matrilineage versus other sources. Officeholdersfallow matrilineage land longer than they do land obtained from othersources, although this difference is statistically significant only for hold-ers of noninherited offices. In contrast to the result in column 3, whendisaggregating we find that holders of inherited office fallow plots longerthan nonofficeholders, even if they are not obtained from the matri-lineage.

We saw in table 1 that individuals with local offices expressed moreconfidence in their rights over their plots, and table 7 shows that theseofficeholders fallow their land much longer than other individuals andthat these variations in fallowing choices associated with local politicalstatus are mostly limited to matrilineage land. The complexity and am-biguity of land rights in the study area were discussed in Section II. Oneconsequence of this complexity is that individuals commonly cultivateplots obtained from a variety of sources and through a variety of ar-rangements. This variety permits us to examine the within-cultivatordeterminants of fallowing behavior. The key advantage of this strategyis that we can distinguish between determinants of fallowing that operateat the individual level, such as the shadow costs of factors of productionor unobserved ability, and those that might operate at the level of theplot#cultivator interaction, such as the security of tenure over a givenplot.

In column 1 of table 8, we show that fallow durations vary across theplots of a given cultivator, depending on the source of the plot (theseare also conditional on spatial fixed effects). The excluded category inthis regression is the set of plots obtained via noncommercial arrange-ments, from individuals who are not close family members.14 As reportedin table 2, somewhat more than one-quarter of plots are obtainedthrough commercial transactions, either fixed-rent or sharecroppingcontracts. These plots are left fallowed for almost 8 months longer than

14 The sample size is smaller in col. 1 than in col. 2 or 3 because for some plots we aremissing data on the identity of the individual from whom the plot was obtained.



TABLE 8Within-Cultivator Fallowing Choices

Dependent Variable: Fallow Duration

All Plots: OLS

ExcludeCommercialPlots: OLS

(1) (2) (3) (4) (5)

Plot in same abusua as cultivator .05(.18)

.07(.22)

"1.10(.37)

.86(.62)

Cultivator holds office#Plot insame abusua as cultivator

2.30(.86)

5.96(2.16)

Cultivator holds inherited of-fice#Plot in same abusua ascultivator

3.49(1.03)

6.57(1.60)

Cultivator holds noninherited of-fice#Plot in same abusua ascultivator

2.28(.54)

4.03(1.15)

Plot obtained commercially .64(.26)

Plot obtained from spouse ".58(.41)

Plot obtained from family .83(.36)

Observations 388 402 402 266 266Fixed effects Individual cultivator and spatial fixed effects

(250 meters)Note.—Standard errors are in parentheses.

other plots farmed by the same cultivator.15 Plots obtained from one’sspouse may be left fallow less than other plots, but the difference is notstatistically significant at conventional levels. Plots that are obtainedfrom other close family members are fallowed for almost 10 monthslonger than plots obtained from individuals who are not related.

There is important variation in fallow durations across the plots cul-tivated by a given individual, depending on the provenance of the plot.This variation corresponds to the confidence that individuals expressregarding their rights in focus group discussions, in which it was arguedthat commercial transactions or close family ties help to secure one’sability to reestablish cultivation on a fallowed plot; whereas women ex-pressed particular concern over their ability to maintain control overplots obtained indirectly from another source via their spouse. Whererelevant, these results also correspond to cultivators’ self-assessed rightsover plots. Farmers claim the right to rent out land obtained from familyon 39 percent of such plots, but claim this right only on 3 percent ofplots obtained from nonfamily and only on 1 percent of plots obtained

15 We find no significant differences in fallowing choices between sharecropped andfixed-rent contracts.



from their spouse. Similar patterns are observed for the right to lendout the plot, sell it, or decide who will inherit it.16

We saw in column 3 of table 7 that, conditional on household fixedeffects, officeholders fallow matrilineage land much longer than theydo land from outside the matrilineage. This accords with the literatureon land rights in southern Ghana, which makes it clear that tenuresecurity is not a universal attribute of an individual. Rather, an indi-vidual’s security of tenure over a particular plot reflects that individual’sposition within the local social and political hierarchy and the mannerin which that plot was obtained.

Looking only across plots cultivated by a given individual, in column2 of table 8 we show that officeholders fallow land from within theirown matrilineage for more than 2 years longer than they do other plotsthat they cultivate. Because officeholders are in a superior political po-sition, they are more confident of their ability to reestablish cultivationon fallowed plots that they have obtained through the matrilineageallocation process and therefore leave such plots fallow for longer.

We replicate these results in column 3 for disaggregated offices: hold-ers of both inherited and noninherited offices fallow their matrilineageplots longer than they do their other plots. The point estimate is largerfor those who hold inherited office than for those who do not, but thedifference is not statistically significant. Officeholders leave their ma-trilineage-obtained plots to fallow for 2–4 years longer than they dotheir other plots.

We expect the increased security of plots cultivated by officeholdersto be particularly evident on plots that were obtained via this politicalprocess. Therefore, we restrict attention in the specification reportedin column 4 to plots that were not obtained through commercial trans-actions. Officeholders fallow noncommercial land from within their ownmatrilineage for almost 6 years longer than they do noncommercialland from other sources; in stark contrast, nonofficeholders fallow non-commercial land from within their own matrilineage even less than theydo land from other sources. In column 5, we again disaggregate officesand find the same general story. In this case, we no longer find thatnonofficeholders leave their matrilineage land fallow less than theirother plots. Holders of both inherited and noninherited offices leavetheir matrilineage land fallow much longer (4–7 years) than their otherplots.

Officeholders leave land fallow for longer periods than other indi-viduals within these villages. However, this is not simply a matter ofofficeholders having a superior political and social position than other

16 No farmer cultivating a plot commercially claims the right to rent it or lend it out,sell it, or decide who will inherit it.



individuals and thus having more tenure security in general. Instead,their political power is exercised within specific contexts. Officeholdersare able to use their social and political status to secure their rights overplots that they obtain through the explicitly political process of landallocation through the matrilineage. However, this ability does not fullyspill over into improved security of tenure in other contexts.

D. Tenure Duration and the Hazard of Expropriation

We have argued that the dramatic variation we observe in investmentacross plots is driven by variation in the likelihood that these plots willbe expropriated while fallow. In this subsection, we provide direct evi-dence of this variation in tenure security and show that it correspondsto the variation in fallowing choices that we observe.

For each plot in our data we have information on the duration oftenure. That is, we know how long the current cultivator has controlledthe plot. The expected duration of tenure depends on , the likelihoodqp

of losing the plot in any year in which it is fallow. Plots that are heldmore securely will, on average, be held for longer durations.17

We have shown that fallowing varies according to officeholding statusand the origin of the plot. These findings correspond to the ethno-graphic evidence on tenure security discussed in Section II, which alsoemphasizes the potential importance of the gender of the cultivator.Suppose, therefore, that , where is a vector that in-′q p exp (Q g) Qp p p

cludes indicators of the gender and officeholding status of the cultivatorof p, an indicator equal to one if the plot belongs to the same matri-lineage as its cultivator, and interactions of these indicator variables.

In column 1 of table 9 we show the mean tenure durations acrossthe categories defined by . As expected, officeholders have held theirQp

plots longer than nonofficeholders, and within each category of indi-vidual, plots that come from within their matrilineage have been heldlonger than plots obtained from other matrilineages (except for femaleofficeholders, where the standard errors are enormous, reflecting thetiny sample of such individuals).

Data on the duration of tenure provide direct evidence on the vari-ation across plots in . Consider a set of plots (say, all plots from withinqp

the matrilineage controlled by officeholding males) with a common. If cultivation were instantaneous, so that in every year the prob-q p kp

17 This statement is subject to the caveat that fallow durations are not so much longeron more securely held plots as to outweigh the direct effect of increased security onaverage tenure durations. See eq. (9).


TAB

LE

9T

he

Haz

ard

ofE

xpro

pria

tion

Year

sPl

otH

asB

een

Hel

dby

Cul

tiva

tor:

Mea

n(1

)

Ann

ual

Rat

eof

Exp

ropr

iati

onW

hil

eFa

llow

:*M

axim

umL

ikel

ihoo

d(2

)

Fall

owD

urat

ion:

Mea

n(3

)

Profi

t:M

ean

(4)

Impl

ied

Dis

coun

tR

ate

Low

erB

ound

(5)

Upp

erB

ound

(6)

Offi

ceho

lder

s:M

ale,

plot

from

sam

eab

usua

16.4

1(1

.85)

.20

(.02

4)4.

9.7

9.0

71.

04

Fem

ale,

plot

from

sam

eab

usua

11.0

9(1

.20)

.24

(.03

4)5.

0.3

7.2

58

Mal

e,pl

otfr

omdi

ffer

ent

abus

ua12

.90

(1.8

1).2

2(.

019)

4.4

.83

.12

1.96

Fem

ale,

plot

from

diff

eren

tab

usua

12.7

4(7

.32)

.27

(.02

9)4.

0"

.62

1N

A

Non

offic

ehol

ders

:M

ale,

plot

from

sam

eab

usua

8.90

(.74

).2

9(.

015)

4.3

.74

.09

4.25

Fem

ale,

plot

from

sam

eab

usua

7.86

(1.2

0).3

4(.

022)

3.7

".4

6N

AN

A

Mal

e,pl

otfr

omdi

ffer

ent

abus

ua6.

59(.

51)

.35

(.01

8)3.

8.5

5.2

58

Fem

ale,

plot

from

diff

eren

tab

usua

4.95

(.79

).4

1(.

023)

4.0

".0

3N

AN

A

Obs

erva

tions

753

753

770

770

NA

NA

Not

e.—

Stan

dard

erro

rsar

ein

pare

nthe

ses.

*C

alcu

late

dfr

omm

axim

umlik

elih

ood

estim

ates

ofg

in(9

)an

dre

port

edin

tabl

eA

1.St

anda

rder

rors

calc

ulat

edfr

omta

ble

A1

usin

gth

ede

ltam

etho

d.



ability of losing the plot is k, the expected average tenure in the crosssection of these plots is

! 1 " ktT p k t(1 " k) p , (8)!k ktp0

which obviously decreases in k.18

In fact, cultivation occurs over a period of 2 years in this fallow system,and during cultivation the probability of losing land drops to zero. Ifit were possible to cultivate continuously without fallowing, land wouldnot be lost at all. Hence the expected value of tenure duration dependson both and the fallow duration . Equation (8) is not correctq t*p p

because it does not take into account the period during which the plotis cultivated. Since plots are not expropriated during the 2-year periodof cultivation, the likelihood of observing a plot of tenure duration dis

1 Nt*pl(d, q , t*) p (1 " q )p p pt* # 2t"2 t"1# {(t " 1)(1 " q ) # 2(1 " q )p p

t# [t* " (t " 1)](1 " q ) }, (9)p p

where is the number of completed fallow-cultivationN p int[d/(t* # 2)]p

cycles associated with duration d given and the remaindert* t p d "p

. The first term is straightforward, being the likelihood of a plot sur-Nviving through N complete fallow cycles, during each of which it is atrisk of being lost for years. The final term reflects the fact that duringt*pevery fallow cycle there are two years during which the plot is beingcultivated, and these two years may occur during any two consecutiveyears of the cycle (because the starting year of the cycle is arbitrary).Given , we can estimate g using the likelihood function′q p exp (Q g)p p

implied by (9). The maximum likelihood estimates are presented inAppendix table A1; the more interesting implied hazard rates areqp

presented in column 2 of table 9.The most striking feature of these results overall is the magnitude of

the hazard of plot loss faced by people in Akwapim. Even male office-holders cultivating plots from within their own matrilineage face a 20percent chance of losing a plot in any year in which it is left fallow. Thisprobability rises to over 40 percent for female nonofficeholders culti-vating plots outside their matrilineage.

This dramatic risk makes it unsurprising that fallows are relatively

18 This calculation assumes that this is a stationary environment. In each period, plotsare lost with probability (p k). Stationarity requires that plots arrive with the sameqp

probability.



short in our sample; leaving a plot fallow entails a striking risk of losingthe plot. Moreover, the doubling of the annual risk of loss dependingon personal and plot characteristics rationalizes the large difference inthe fallowing choices that we observe across the plots in our sample.

E. How Inefficient Is Fallowing in Ghana?

In a fully efficient allocation within a village, fallow durations would bethe same on all similar plots (by eq. [2], noting that shadow prices arethe same across plots within an efficient allocation). Even if shadowprices vary across households because of other imperfections in factormarkets, similar plots within a household should be fallowed similarly.However, we have shown that fallow durations on similar plots varywithin households, and even across the plots held by an individual. Intable 9 we have shown that the annual hazard that an individual willlose a plot while it is fallow depends on the position of the individualin local political and social hierarchies and the manner in which theplot was acquired. In this subsection we provide a rough estimate of theproductivity costs of the inefficient fallowing that results from insecurityof land tenure.

The linear approximation to the profit function presented in table 4implies that per-hectare profits can be increased without limit for suf-ficiently long fallow periods. Since this is not possible, we now estimatea profit function that is potentially concave in fallow duration. We wouldlike to estimate a semiparametric (say, partial linear) model that placesfew restrictions on the relationship between profit and . However, ourtp

data do not contain sufficient information to detect the degree of con-cavity of the profit function without additional aid.

We observed in Section II that soil scientists working in the regionconclude that a fallow duration of 6–8 years is sufficient to maintainsoil fertility. Therefore, we impose the restriction that fallow durationsof longer than 7 years have no further impact on profits, and we specifya flexible functional relationship between fallow duration and profits.We estimate the profit function

p p X b # g(t ) # l # e , (10)pt p p h ,t ptp

where the contribution of fallow duration ( ) to profits istp

aa ln (t # b) " t for t ≤ 7p p p7 # bg(t ) p (11)p a{a ln (7 # b) " 7 for t 1 7.p7 # b

The values a and b determine the slope and concavity of the relationship



Fig. 2.—Nonlinear IV estimate of the profit function. Central line: , whereˆˆ ˆg(a, b) (a,are point estimates from nonlinear IV estimates of the profit function. Outer band: 90b)

percent confidence interval for and corresponding values of . Inner band: 90 percentˆa bconfidence interval for and corresponding values of . Confidence intervals are takenˆ ˆb afrom 1,000 block-bootstrap iterations, with blocks defined by household-year groups.

between fallowing and profits. The second term simply ensures that thederivative of the function is zero at . This profit function is esti-t p 7p

mated by nonlinear IV, with the same instrument set that was employedin table 4.19 The results are reported in figure 2. Before we proceed,one point should be made clear: these are wide confidence bounds. Wecan be confident that is upward sloping and concave, but theˆˆg(a, b)data are not sufficiently rich to provide us with a tight estimate of thedegree of concavity. With that caveat in mind, we now have in hand therequirements for two sets of calculations of interest. The first set involvescalibrating the output lost to inefficient fallowing behavior. The secondconcerns the household-specific discount rates that rationalize the cho-

19 Define and let be the value of the g function at parameter values′b p (b, a, b) g (a, b)pt

a and b for plot pt. Let be the household-year average profit of the household thatppt

cultivates plot p, and similarly for and and the instruments . We define the within-¯X g Zptp p

household differences , , , andd d d dp p p " p X p X "X Z p Z "Z g (a, b) p g (a,pt pt pt pt p p p pt ptp p

. Then if¯b) " g (a, b)pt

d d d d˜u (b) p p " X b " g (a, b),pt pt pt pt

our estimate minimizes the quadratic form′ d d ′ d "1 d ′˜ ˜u(b) Z (Z Z ) Z u(b).

Ninety percent confidence intervals are constructed using 1,000 bootstrap iterations (clus-tered at the household-year level).



sen fallow durations, given the estimated profit function and hazardsof losing land while it is fallow.

Suppose that all farmers adjusted their fallow durations to the meanfallow duration within their household. This experiment would elimi-nate the inefficiency associated with the intrahousehold dispersion infallow durations that is associated with variations in tenure security acrossplots within the household, but of course does not account for cross-household variations in security of tenure. The change in profits forhousehold h is

2 2(p # D ) # " p # , (12)! p p pe2 # t 2 # tp!Ph p

where

a ae eD { a ln (t # b) " t " a ln (t # b) " t .p p p[ ] [ ]7 # b 7 # b

The term is the “experimental” fallow duration, here equal to theetaverage duration of fallow on plots held by household h; is the ab-Dp

solute change in the level of profits on each household from this change;and is the proportion of years the plot would be cultivated2/(2 # t )h

given the change in fallow duration. The average (median) change inprofits per household associated with this change, given our estimatesof a and b, is approximately 60,000 (0) cedis, compared to average(median) household farm profits of 240,000 (0) cedis.20 This calculationabstracts from any cross-household variation in tenure security and thusprovides a lower bound to the change in profits associated with moresecure property rights.

An alternative would be to consider the implications of moving allplots to a fallow duration that corresponds to the mean duration weobserve on plots that are cultivated by officeholders on plots that theyobtain from their own matrilineage. This average is 5 years, so we repeatthe calculation above with set to 5. In this case, the average (median)etchange in household farm profits is 195,000 (75,000) cedis. This is likelyto be an overestimate because it assumes that the discount rate is equalacross households, which is unlikely to be correct in an environmentof highly imperfect capital markets.

A speculative calculation can help to put these numbers into a broaderperspective. Approximately 434,000 hectares of Ghana’s farmland areplanted to maize and cassava and located in regions where we might

20 For reference, the mean (median) value of household farm profits without deductingthe imputed value of household labor used on plots is 665,000 (320,000) cedis.



expect the land tenure system to be similar.21 If the yield losses frominefficient fallowing are similar on all of this land, then we estimate theaggregate costs at 86 billion cedis. This translates into just under 1percent of 1997 national GDP.22 Another perspective on this magnitudeis provided by the depth of poverty in Ghana. The aggregate yield lossin these four regions is approximately 6 percent of the national povertygap.23

Finally, we calculate the discount rates that rationalize the observedfallowing choices, given our estimates of the profit function and hazardsof losing land while fallow. If households are risk neutral, then thedecision to fallow a plot for years impliestp

1 " qpp(t ) ≥ p(t # 1) (13)p p( )1 # rh

and

1 " qpp(t ) ≥ p(t " 1), (14)p p( )1 # rh

which together imply

ln (1 " q ) # lnp(t ) " lnp(t " 1) ≥ ln (1 # r )p p p h

≥ ln (1 " q ) # lnp(t # 1) " lnp(t ).p p p

(15)

We present these bounds for the mean values of tenure duration andprofits achieved in each of our broad categories of tenure security intable 9.24 Because of the sharp concavity of , the bounds are wide,p(t )p

but the key point is clear: the high hazard of losing a plot while it isfallow implies that the short mean fallowing decisions we observe are

21 The regional breakdown of farming area comes from “Special Report: FAO/WFPCrop and Food Supply Assessment Mission to Northern Ghana,” World Food Programme,Food and Agricultural Organization (March 13, 2002). We use area planted to maize andcassava figures from 2000 for the western, central, eastern, and Ashanti regions. As perpersonal communication with the Ministry of Food and Agriculture, we use the larger ofmaize or cassava area figures to account for intercropping (this biases the area figuredownward since it excludes some single cropped fields).

22 GDP figures come from the World Bank’s World Development Indicators database.23 The poverty gap is the amount that, if perfectly targeted, would bring all the poor

to the poverty line. From 1998 national household survey data (the Ghana Living StandardsSurvey, round 4), the poverty gap is estimated at 14 percent of the poverty line. We canuse this figure to calculate the aggregate poverty gap, which is about 1.55 trillion cedis(converting the 1998 cedis to 1997). This is based on a poverty line of 688,401 cedis percapita. We are grateful to Kalpana Mehra for these statistics.

24 We cannot calculate the bounds for plots cultivated by women who do not hold officebecause mean profits in these plots are negative at the baseline fallowing duration.



consistent with reasonably low discount rates, in the range of 10–30percent per year.

V. Conclusion

We find that insecure land tenure in Ghana is associated with greatlyreduced investment in land fertility. Individuals who are not central tothe networks of social and political power that permeate these villagesare much more likely to have their land expropriated while it is fallow.Their reduced confidence of maintaining their rights over land whileit is fallow induces such individuals to fallow their land less than wouldbe technically optimal. As a consequence, farm productivity for theseindividuals is correspondingly reduced. There is a strong gender di-mension to this pattern since women are rarely in positions of sufficientpolitical power to be confident of their rights to land. So women fallowtheir plots less than their husbands and achieve much lower yields.

These large effects of land tenure insecurity on investment and pro-ductivity stand in contrast to the great majority of the recent microeco-nomic literature on property rights and investment. That literature tendsto find no or only subtle impacts of insecure property rights on invest-ment behavior (e.g., Besley 1998; Brasselle et al. 2002; Jacoby et al. 2002;Galiani and Schargrodsky 2006; Field 2007). The large effects that wefind of tenure insecurity are likely a consequence of three factors. First,the degree of insecurity in property rights that we document is huge.Individuals have on the order of a one in three chance of losing controlover a plot in any year in which it is not cultivated. Expropriation riskis therefore a very salient aspect of the economic environment. More-over, there is very large variation across the sample plots in the extentof tenure insecurity. The annual hazard of losing a plot while fallowapproximately doubles between the least and most securely held plots.Second, many studies focus on de jure rights over land. De jure variationin the security of property rights may not be reflected in variations inde facto tenure security. Third, we study a well-measured and highlyproductive investment that everyone undertakes. We therefore avoidsome of the econometric challenges that many researchers face.

Our results provide support for the argument that in West Africa “theprocess of acquiring and defending rights in land is inherently a politicalprocess based on power relations among members of the social group.. . . A person’s status . . . can and often does determine his or hercapacity to engage in tenure building” (Bassett and Crummey 1993, 20).Rights over a particular plot of land are political: they depend on thefarmer’s ability to mobilize support for her right over that particularplot. Hence the security of tenure is highly dependent on the individ-ual’s position in relevant political and social hierarchies. Even condi-



tional on the individual’s position, her security depends on the circum-stances through which she came to obtain access to the particular plot.

The lack of success of widespread land titling programs in Africa hasled many to question the conventional wisdom regarding the impor-tance of secure property rights for investment in land. Bassett (1993,4) notes that “colonial administrators, African elites, and foreign aiddonors have historically viewed indigenous landholding systems as ob-stacles to increasing agricultural output. . . . There is a need to tran-scend [the World Bank’s] technocratic and theological approaches thatposit a direct link between freehold tenure and productivity.” On thebasis of her rich understanding of Akan land tenure, Berry (2001, 155–56) argues that “contrary to recent literature, which argues that sus-tainable development will not take place unless rights to valuable re-sources are ‘clearly defined, complete, enforced and transferable,’ assetsand relationships in Kumawu appear to be flexible and resilient becausethey are not clearly defined, or completely and unambiguously trans-ferable.”25

We have shown that a great deal of potential output is lost in thestudy area because land tenure is insecure. Pande and Udry (2006)provide a summary of the historical origins of the institution in whichland use rights are allocated through the matrilineage. They show thatthis institution emerged during a long period of land abundance, duringwhich fallow periods on virtually all land were sufficiently long for fullrestoration of land fertility. Tenure insecurity would have no conse-quence for fallow durations under such conditions. However, over thepast several decades land has become more scarce, and therefore in-dividuals’ uncertainty regarding their ability to reestablish cultivationafter a period of fallow now has implications for fallow durations andhence productivity. We do not adhere to a view that institutions nec-essarily adjust to capture all potential Pareto gains. However, the per-sistence of this method of land allocation in the face of the losses ofoutput associated with tenure insecurity requires investigation.

We interpret the resilience of this system of land tenure to its crucialand flexible role in redistributing resources in the face of unobservedvariations in need. Similar processes of land reallocation through cor-porate groups exist in most societies in West Africa; as a consequence,the region is distinguished by the almost complete absence of a rurallandless class. This system may provide important insurance in times ofneed and a remarkable degree of social stability due to the redistributionof land within rural communities. This paper reveals that this stability

25 Doubt is not confined to those studying property rights in Africa. Regarding Europeangrowth, Clark (2007, 727) argues that “quantitative research in recent years suggests thatcommon rights, at least by the seventeenth century, had negligible impacts on agriculturalperformance” (citing Allen [1982], Hoffman [1988], and Clark [1998]).



and insurance come at a steep price paid by those distant from localcenters of political power.

Appendix

TABLE A1Determinants of the Rate of Expropriation

Maximum Likelihood:Annual Hazard

of ExpropriationWhile Fallow*

EstimateStandard

Error

1 if officeholder ".66 .141 if woman .27 .10Plot in same abusua as cultivator ".29 .10Officeholder#Plot in same abusua as cultivator .16 .21Constant ".62 .08Observations 753

* Maximum likelihood estimates of g, where annual rate of expropriation , and the likelihood′q p exp (Q g)p

is defined in eq. (9).

References

Acemoglu, D., S. Johnson, and J. A. Robinson. 2001. “The Colonial Origins ofComparative Development: An Empirical Investigation.” A.E.R. 91 (5): 1369–1401.

———. 2002. “Reversal of Fortune: Geography and Institutions in the Makingof the Modern World Income Distribution.” Q.J.E. 117 (4): 1231–94.

Ahn, P. M. 1979. “The Optimum Length of Planned Fallows.” In Soils Researchin Agroforestry, edited by H. Mongi and P. Huxley. Nairobi: Internat. CentreRes. Agroforestry.

Akresh, Richard. 2005. “Understanding Pareto Inefficient Intrahousehold Al-locations.” IZA Discussion Paper no. 1858, Inst. Study Labor, Bonn.

Allen, R. C. 1982. “The Efficiency and Distributional Consequences of Eigh-teenth Century Enclosures.” Econ. J. 92: 937–53.

Amanor, K. S. 1994. The New Frontier: Farmers’ Response to Land Degradation; a WestAfrican Study. London: Zed Books.

———. 2001. Land, Labour and the Family in Southern Ghana: A Critique of LandPolicy under Neo-liberalism. Uppsala: Nordiska Afrikainstutet.

Austin, G. 2004. Labour, Land, and Capital in Ghana: From Slavery to Free Labourin Asante, 1807–1956. Rochester, NY: Univ. Rochester Press.

Banerjee, A. V., P. J. Gertler, and M. Ghatak. 2002. “Empowerment and Efficiency:Tenancy Reform in West Bengal.” J.P.E. 110 (2): 239–80.

Bassett, T. 1993. “Cartography, Ideology and Power: The World Bank in NorthernCote d’Ivoire.” Passages: A Chronicle of the Humanities, no. 5: 8–9. http://quod.lib.umich.edu/p/passages/.

Bassett, T., and D. Crummey. 1993. Land in African Agrarian Systems. Madison:Univ. Wisconsin Press.



Benneh, G. 1988. “The Land Tenure and Agrarian System in the New CocoaFrontier: Wassa Akropong Case Study.” In Agricultural Expansion and PioneerSettlements in the Humid Tropics, edited by W. Manshard and W. B. Morgan.Tokyo: United Nations Univ.

Berry, S. 1993. No Condition Is Permanent: The Social Dynamics of Agrarian Changein Sub-Saharan Africa. Madison: Univ. Wisconsin Press.

———. 2001. Chiefs Know Their Boundaries: Essays on Property, Power and the Pastin Asante, 1896–1996. Portsmouth, NH: Heinemann.

Besley, T. 1995. “Property Rights and Investment Incentives: Theory and Evi-dence from Ghana.” J.P.E. 103 (5): 903–37.

———. 1998. “Investment Incentives and Property Rights.” In New PalgraveDictionary of Economics and the Law, edited by P. Newman. New York: PalgraveMacmillan.

Biebuyck, D. P. 1963. African Agrarian Systems: Studies Presented and Discussed.London: Oxford Univ. Press (for Internat. African Inst.).

Binswanger, H. P., K. Deininger, and G. Feder. 1995. “Power, Distortions, Revoltand Reform in Agricultural Land Relations.” In Handbook of Development Eco-nomics, vol. 3, edited by J. Behrman and T. N. Srinivasan. Amsterdam: ElsevierSci.

Brasselle, A.-S., F. Gaspart, and J.-P. Platteau. 2002. “Land Tenure Security andInvestment Incentives: Puzzling Evidence from Burkina Faso.” J. DevelopmentEcon. 68 (2): 373–418.

Bromley, D. 1989. “Property Relations and Economic Development: The OtherLand Reform.” World Development 17 (6): 867–77.

Bruce, J. W., and S. E. Migot-Adholla, eds. 1994. Searching for Land Tenure Securityin Africa. Dubuque, IA: Kendall/Hunt.

Chiappori, P. 1988. “Rational Household Labor Supply.” Econometrica 56 (1): 63–89.

Clark, G. 1998. “Commons Sense: Common Property Rights, Efficiency, andInstitutional Change.” J. Econ. Hist. 58 (March): 73–102.

———. 2007. “The Long March of History: Farm Wages, Population and Eco-nomic Growth, England 1209–1869.” Econ. Hist. Rev. 60 (February): 97–136.

Conley, T. G. 1999. “GMM Estimation with Cross Sectional Dependence.” J.Econometrics 92 (1): 1–45.

de Rouw, A. 1995. “The Fallow Period as a Weed-Break in Shifting Cultivation(Tropical Wet Forests).” Agriculture, Ecosystems and Environment 54 (1–2): 31–43.

De Soto, H. 2000. The Mystery of Capital: Why Capitalism Triumphs in the West andFails Everywhere Else. New York: Basic Books.

Easterly, W., and R. Levine. 2003. “Tropics, Germs, and Crops: How EndowmentsInfluence Economic Development.” J. Monetary Econ. 50 (1): 3–39.

Engerman, S., and K. Sokoloff. 2003. “Institutional and Non-institutional Ex-planations of Economic Differences.” In Handbook of New Institutional Eco-nomics, edited by Claude Menard and Mary M. Shirley. Dordrecht: Springer.

Faustmann, M. [1849] 1995. “Calculation of the Value Which Forest Land andImmature Stands Possess for Forestry.” In Economics of Forestry, edited by R. A.Sedjo. International Library of Environmental Economics and Policy. Alder-shot, UK: Ashgate.

Field, E. 2007. “Entitled to Work: Urban Property Rights and Labor Supply inPeru.” Q.J.E. 122 (4): 1561–1602.

Firmin-Sellers, K. 1996. The Transformation of Property Rights in the Gold Coast: An



Empirical Analysis Applying Rational Choice Theory. New York: Cambridge Univ.Press.

Fred-Mensah, B. K. 1996. “Changes, Ambiguities and Conflicts in Buem, EasternGhana.” Manuscript, Johns Hopkins Univ.

Galiani, Sebastian, and Ernesto Schargrodsky. 2006. “Property Rights for thePoor: Effects of Land Titling.” Manuscript, Univ. Torcuato Di Tella.

Gine, X. 2005. “Land Security in Rural Thailand: Evidence from a PropertyRights Reform.” Manuscript, World Bank, Washington, DC.

Goldstein, M., and C. Udry. 1999. “Agricultural Innovation and Resource Man-agement in Ghana.” Manuscript, Internat. Food Policy Res. Inst., Washington,DC, http://www.ifpri.org/training/gender/udry.pdf.

Hall, R. E., and C. I. Jones. 1999. “Why Do Some Countries Produce So MuchMore Output per Worker than Others?” Q.J.E. 114 (1): 83–116.

Hoffman, P. 1988. “Institutions and Agriculture in Old Regime France.” Politicsand Society 16 (2–3): 241–64.

Jacoby, Hanan, Scott Rozelle, and G. Li. 2002. “Hazards of Expropriation: TenureInsecurity and Investment in Rural China.” A.E.R. 92 (December): 1420–47.

Johnson, S., J. McMillan, and C. Woodruff. 2002. “Property Rights Finance.”Working Paper no. W8852 (March), NBER, Cambridge, MA, http://ssrn.com/abstractp305078.

Jones, Christine. 1986. “Intra-household Bargaining in Response to the Intro-duction of New Crops: A Case Study from North Cameroon.” In UnderstandingAfrica’s Rural Households and Farming Systems, edited by Joyce Lewinger Moock.Boulder, CO: Westview.

Klingelhofer, A. 1972. “Agriculture.” In Akwapim Handbook, edited by D. Bro-kensha. Accra: Ghana Pub. Corp.

Krishna, R. 1964. “Theory of the Firm: Rapporteur’s Report.” Indian Econ. J. 11:514–25.

McCaskie, T. C. 1995. State and Society in Pre-colonial Asante. New York: CambridgeUniv. Press.

Mokyr, J. 2002. The Gifts of Athena: Historical Origins of the Knowledge Economy.Princeton, NJ: Princeton Univ. Press.

Nweke, Felix. 2004. “New Challenges in the Cassava Transformation in Nigeriaand Ghana.” EPTD Discussion Paper no. 118, Internat. Food Policy Res. Inst.,Washington, DC.

Nweke, Felix, D. Spencer, and J. Lynam. 2002. The Cassava Transformation: Africa’sBest Kept Secret. Lansing: Michigan State Univ. Press.

North, D. C. 1981. Structure and Change in Economic History. New York: Norton.Ollenu, N. A. 1962. Principles of Customary Law in Ghana. Law in Africa Series,

no. 2. London: Sweet & Maxwell.Otsuka, K., A. Quisumbing, E. Payongayong, and J. Aidoo. 2003. “Land Tenure

and the Management of Land and Trees: The Case of Customary Land TenureAreas of Ghana.” Environment and Development Econ. 8 (1): 77–104.

Pande, R., and C. Udry. 2006. “Institutions and Development: A View fromBelow.” In Advances in Economics and Econometrics: Theory and Applications, NinthWorld Congress, edited by Richard Blundell, Whitney Newey, and Torsten Pers-son. New York: Cambridge Univ. Press.

Peters, P. E. 1994. Dividing the Commons: Politics, Policy, and Culture in Botswana.Charlottesville: Univ. Press Virginia.

Place, F., and P. Hazell. 1993. “Productivity Effects of Indigenous Tenure Systemsin Sub-Saharan Africa.” American J. Agricultural Econ. 75 (1): 10–19.

Rattray, R. S. 1923. Ashanti. Oxford: Clarendon.



Ravallion, M., and Dominique Van de Walle. 2008. “Does Rising LandlessnessSignal Success or Failure for Vietnam’s Agrarian Transition?” J. DevelopmentEcon. 87 (October): 191–209.

Sawadogo, Jean-Pierre and Volker Stramm. 2000. “Local Perceptions of Indig-enous Land Tenure Systems: Views of Peasants, Women and Dignitaries in aRural Province of Burkina Faso.” J. Modern African Studies 38 (2): 279–94.

Singh, I., L. Squire, and J. Strauss. 1986. Agricultural Household Models: Extensions,Applications, and Policy. Baltimore: Johns Hopkins Univ. Press.

Smith, A. 1974. An Inquiry into the Nature and Causes of the Wealth of Nations.London: Penguin.

Udry, C. 1996. “Gender, Agricultural Productivity, and the Theory of the House-hold.” J.P.E. 104 (5): 1010–46.

Wilks, I. 1993. Forests of Gold: Essays on the Akan and the Kingdom of Assante. Athens:Ohio Univ. Press.


The Profits of Power: Land Rights and Agricultural ...

Documents

Transcript of The Profits of Power: Land Rights and Agricultural ...